The IEEE 802.3 specified twisted pair Ethernet physical interfaces 10Base-T (10 Mbps “Ethernet”), 100Base-Tx (100 Mbps “Fast Ethernet”), and 1000Base-T (1000 Mbps “Gigabit Ethernet”) are well established and widely deployed technologies supporting high speed computer internetworking in local area networks. The popularity of these interfaces is related to the abundance of low cost and easily installed structured cabling components utilized to provide connectivity between networked devices.
1000Base-T, and to a lesser extent, 100Base-Tx, employ sophisticated physical (electrical) layer signaling schemes that require high speed, real-time digital signal processing technology to produce and to decode signals transmitted on twisted pair cabling. Additionally, all 10/100/1000Base-T interfaces rely on isolation and common mode suppression transformers and high bandwidth, impedance matched circuit connections.
While data transmission performance of a network interface is very much dependent upon the integrity and performance of each 10/100/1000Base-T interface (port) at the physical layer, the relationship is typically obscured by many other factors such as connection distance, cabling and connection environments, physical layer error correction capability (1000Base-T), and internet packet re-transmission protocols at higher network layers.
Assessing the physical layer performance characteristics of a 10/100/1000Base-T port independent of these other factors conventionally requires very specialized test setups and measurements that are historically very distinct from ordinary connection and usage of that interface. More specifically, IEEE 802.3 specifications dictate test methods involving high speed digital oscilloscopes, high bandwidth active differential probes, specialized measurement interfaces or fixtures, specialized transceiver device test modes, as well as other forms of electronic test equipment. IEEE 802.3 test methods are focused mainly on transmitter and interface characteristics and are less specific about receiver testing. The testing described is typically performed on a single transmitting wire pair meaning that two independent sets of test data (MDI, MDI-X) must be gathered for 100Base-Tx interfaces and four independent sets of test data (Pairs A, B, C, and D) must be gathered for 1000Base-T interfaces. High speed oscilloscope measurements performed in the time domain require careful attention to interface fixtures, oscilloscope probe characteristics and calibrations, and to oscilloscope channel calibrations.
Several producers of high speed digital oscilloscopes do offer semi-automated “scope-ware” solutions and pre-fabricated test fixtures to facilitate testing of 100Base-Tx and 1000Base-T transmitted signal characteristics. These solutions help with the oscilloscope configuration and test data collection aspects of certain 802.3 measurements. However, they do not address core metrology issues such as probe characterization, probe accuracy, fixture calibration, and calibration references. Users of those systems must address all of those issues of absolute measurement accuracy independently.
Because of the cost and complexity of traditional physical layer testing methods and solutions, many producers of products with one or more Ethernet ports rely on functional link verification or packet transmission verification types of testing. These methods offer plug-and-play connectivity and are much easier to perform. However, they offer very limited parametric insight meaning they cannot assure tested devices will perform properly under all network interface conditions.
As a general matter in digital communication systems, the assessment of receiver performance is a challenging task owing to the need to precisely simulate incoming signals while capturing the outcome of receiver decisions through a direct interface to the decision making entity. Signal simulation is important because receiver testing is ideally carried out with signals that range to the specified input tolerances of a receiver in areas such as signal amplitude, frequency and phase response characteristics, noise content, symbol timing variation, and so on. A well-known digital receiver measurement, bit error ratio or BER, then characterizes the probability of the receiver misinterpreting information at the most fundamental level, that is, the level of bits in a bit stream.
In a prolific and integrated technology such as 10/100/1000Base-T Ethernet, receivers are generally embedded within highly integrated transceiver devices where available information conveying receiver decision performance is not directly available. Consequently, measurements of receiver performance are carried out at the Ethernet packet level where packets that carry many hundreds of bits are deemed either valid or erred in reference to check sum values that are also embedded in those same packets. Erred packets thus create the ambiguity of one to many possible bit errors per packet. Packet flow testing is often restricted to devices with two or more bridged Ethernet ports since packets flowing into the receiver-under-test must then be forwarded back to a packet counting device using a second Ethernet port.
Similarly, signal simulation is impractical in the testing of 100Base-T and 1000Base-T receivers owing to the fact that these are not simple binary signals and that they must comply with complex physical layer signaling protocols in order to establish and maintain a link with another Ethernet interface. Furthermore, commercially available solutions for controlled signal degradation or impairment applicable to 10/100/1000Base-T technology are, rare, poorly characterized, limited in function, and/or expensive.
The task of parametrically testing twisted pair Ethernet interfaces such as 100Base-Tx and 1000Base-T at the physical or electrical signaling, layer using conventional means has been expensive, laborious, and invasive, often defying highly automated approaches. For this reason, the common practice in the network equipment industry has been to rely solely on functional “go/no-go” and functional packet loss testing to qualify interface performance.